Purpose: To study the effect of regional nodal status on predictors of treatment response, failure patterns, and the time-dependent nature of the various pattern of relapse via a hazard function analysis.
Methods and materials: We reviews tumor control data of 496 patients with nasopharyngeal carcinoma (NPC) to whom a radical course of radiotherapy (RT) with or without induction chemotherapy (CT) was given. All alive patients had a median follow-up period of 131 months. Primary tumor (T) and nodal (N) status were staged according to the TNM system of the American Joint Committee. Remote after-loading brachytherapy may be added to teletherapy in T1-2 lesions while induction CT could be given for N3 and/or T4 lesions. Hazard function analysis over 1-year interval was carried out for locoregional or distant relapse.
Results: T stage and brachytherapy were two independent predictors for complete response (CR) at the primary site irrespective of nodal status, whereas N stage and brachytherapy are major determinants for regional CR in node (+) patients. Multivariate analysis revealed that contributors to a relatively long disease-free interval in (1) node (-) patients were for locoregional relapse, induction CT(-) (p = 0.0062) or brachytherapy (+) (p = 0.0268) and for distant relapse, none; (2) node (+) patients were for locoregional relapse, early T stage (p = 0.0377) or regional CR (p = 0.0075) and for distant relapse, induction CT(-) (p = 0.0001) or regional CR (p = 0.0001). In node (-) or (+) patients, primary CR rate yield no independent prognostic value on various types of disease-free survival. Hazard function analysis for relapse revealed that hazard rates are in general negatively correlated with time, being highest at the first year post-treatment, decreasing from time to time, and approaching zero after a longer follow-up period in patients with locoregional CR than in patients without.
Conclusion: Nodal status had no significant impact on predictors of primary CR, whereas in node (+) patients regional CR rate had an independent value in predicting disease-free survival to locoregional and distant relapse. Hazard function analysis revealed a decreasing hazard rate over a protracted post-treatment time in primary and regional CR patients. This indicates the continued risk of late recurrence in this subset of patients for whom long-term observation is recommended.